Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have the following identities

  • $$\sum_{n \le x} \varphi(n) = \frac{1}{2} \sum_{n \le x} \mu(n) \left[\frac{x}{n}\right]^2 + \frac{1}{2}$$
  • $$\sum_{n \le x} \frac{\varphi(n)}{n} = \sum_{n \le x} \frac{\mu(n)}{n} \left[\frac{x}{n}\right]$$

how can I discover a similar one for $$\sum_{n \le x} \frac{\varphi(n)}{n^2}?$$

The reason for changing them from $\varphi$ to $\mu$ is that they are easier to asymptotic analysis. Proving both required the identity $\varphi(n) = \sum_{d|n} \mu(d) \frac{n}{d}$ as the first step so I am guessing that is the first step for this too, but when I try that I get $\sum_{n \le x} \sum_{d|n} \frac{\mu(d)}{nd}$ which doesn't look like a convolution or anything simple like that.

share|cite|improve this question
Need clarification on what valued $d$ take in your sums. – Thomas Andrews May 9 '11 at 19:35
up vote 6 down vote accepted

You are on the correct path, it is very similar to the previous one. As you noted, we have $$\sum_{n\leq x}\frac{\phi(n)}{n^{2}}=\sum_{n\leq x}\frac{1}{n}\sum_{d|n}\frac{\mu(d)}{d}.$$ Switching the order of summation yields $$\sum_{d\leq x}\frac{\mu(d)}{d}\sum_{r\leq\frac{x}{d}}\frac{1}{rd}=\sum_{d\leq x}\frac{\mu(d)}{d^{2}}\sum_{r\leq \frac{x}{d}}\frac{1}{r}=\sum_{d\leq x}\frac{\mu(d)}{d^{2}}H_{\left[\frac{x}{d}\right]}.$$

Where $H_m$ is the $m^{th}$ harmonic number. Now this is possible to evaluate since $$\sum_{n\leq x}\frac{1}{n}=\log x +\gamma+O\left(\frac{1}{x}\right).$$ We then get $$\sum_{d\leq x}\frac{\phi(d)}{d^2}=\sum_{d\leq x}\frac{\mu(d)}{d^{2}}\left(\log\frac{x}{d}+\gamma+O\left(\frac{d}{x}\right)\right).$$

We may rewrite this sum as

$$\log x\sum_{d\leq x}\frac{\mu(d)}{d^{2}}-\sum_{d\leq x}\frac{\mu(d)\log d}{d^{2}}+\gamma\sum_{d\leq x}\frac{\mu(d)}{d^{2}}+O\left(\frac{1}{x}\sum_{d\leq x}\frac{|\mu(d)|}{d}\right)$$

Since $\sum_{d> x}\frac{\mu(d)\log d}{d^2}=O(\log x/x)$, and $\sum_{d\leq x}\frac{\mu (d)}{d}=O(\log x)$, we see that

$$\sum_{n\leq x}\frac{\phi(n)}{n^2}=\log x \sum_{d=1}^\infty \frac{\mu(d)}{d^2}+\gamma \sum_{d=1}^\infty \frac{\mu(d)}{d^2}-\sum_{d=1}^\infty \frac{\mu(d)\log d}{d^2}+O\left(\frac{\log x}{x}\right).$$

Looking at these series, $\sum_{d=1}^\infty \frac{\mu(d)}{d^2}=\frac{1}{\zeta(2)}$ and $\sum_{d=1}^\infty \frac{\mu(d)\log d}{d^2}=\frac{\zeta^{'}(2)}{\zeta(2)^2}$, and so

$$\sum_{n\leq x}\frac{\phi(n)}{n^2}=\frac{\log x}{\zeta(2)}+\frac{\gamma}{\zeta(2)} -\frac{\zeta^{'}(2)}{\zeta(2)^2}+O\left(\frac{\log x}{x}\right).$$

Hope that helps,

Edit: Fixed error with the constant term. I originally missed the $-\log d$ term.

Edit 2: Changed the bound $\sum_{d> x}\frac{\mu(d)\log d}{d^2}=O(1/x)$ to $\sum_{d> x}\frac{\mu(d)\log d}{d^2}=O(\log x/x).$ While the previous one is also true, it requires the prime number theorem. For this result the $\log x$ does not change anything, and there is a much simpler proof. See this answer.

share|cite|improve this answer
Thanks very much! It helps a lot! – quanta May 9 '11 at 22:02

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.